2. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Lecture 16 of 42 Knowledge Engineering and Ontology Engineering Discussion: Description Logics William H. Hsu Department of Computing and Information Sciences, KSU KSOL course page: http://snipurl.com/v9v3http://snipurl.com/v9v3 Course web site: http://www.kddresearch.org/Courses/CIS730http://www.kddresearch.org/Courses/CIS730 Instructor home page: http://www.cis.ksu.edu/~bhsuhttp://www.cis.ksu.edu/~bhsu Reading for Next Class: Sections 10.1 – 10.2, p. 320 – 327, Russell & Norvig 2 nd edition http://en.wikipedia.org/wiki/Ontology_(information_science)

3. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Lecture Outline Reading for Next Class: Sections 10.1 – 10.2 (p. 272 – 319), R&N 2 e Last Class: Resolution Theorem Proving, 9.5 (p. 275-294), R&N 2 e  Proof example in detail  Paramodulation and demodulation  Resolution strategies: unit, linear, input, set of support  FOL and computability: complements (different difficulty) and duals (same)  Theoretical foundations and ramifications of decidability results Today: Prolog in Brief, Knowledge Engineering (KE), Ontologies  Prolog examples  Introduction to ontologies  Description logics and the Web Ontology Language (OWL)  Ontologies defined and ontology design Next Class: More Ontology Design; Situation Calculus Revisited  Knowledge engineering (KE) and knowledge management  KR and reasoning about states, actions, properties Coming Week: Ontologies, Description Logics, Semantic Nets

4. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Acknowledgements © 2006 Horrocks, I. Oxford University (formerly University of Manchester) http://en.wikipedia.org/wiki/Ian_Horrocks Professor Ian Horrocks Professor of Computer Science Oxford University Computing Laboratory Fellow, Oriel College © 2004-2005 Russell, S. J. University of California, Berkeley http://www.eecs.berkeley.edu/~russell/ Norvig, P. http://norvig.com/ Slides from: http://aima.eecs.berkeley.edu

5. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Logic Programming (Prolog) Systems: Review Based on slide © 2004 S. Russell & P. Norvig. Reused with permission.

6. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Prolog Examples in Depth: Review Adapted from slide © 2004 S. Russell & P. Norvig. Reused with permission.

7. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Unit Preference  Idea: Prefer inferences that produce shorter sentences  Compare: Occam’s Razor  How? Prefer unit clause (single-literal) resolvents (α  β with  β  α)  Reason: trying to produce a short sentence (   True  False) Input Resolution  Idea: “diagonal” proof (proof “list” instead of proof tree)  Every resolution combines some input sentence with some other sentence  Input sentence: in original KB or query Unit and Input Resolution: Review Unit resolution Input resolutions

8. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Linear Resolution  Generalization of input resolution  Include any ancestor in proof tree to be used Set of Support (SoS)  Idea: try to eliminate some potential resolutions  Prevention as opposed to cure  How?  Maintain set SoS of resolution results  Always take one resolvent from it  Caveat: need right choice for SoS to ensure completeness Linear Resolution and Set-of-Support: Review Linear resolutions

9. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Subsumption  Idea: eliminate sentences that sentences that are more specific than others  e.g., P(x) subsumes P(A) Putting It All Together Subsumption: Review

10. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence L FOL-VALID (written L VALID ): Language of Valid Sentences (Tautologies) Deciding Membership  Given: KB, α  Decide: KB ⊨ α? (Is α valid? Is ¬α contradictory, i.e., unsatisfiable?) Procedure  Test whether KB  {¬α} ⊢ RESOLUTION   Answer YES if it does L FOL-SAT C (written L SAT C ) Language of Unsatisfiable Sentences Dual Problems Semi-Decidable: L VALID, L SAT C  RE \ REC (“Find A Contradiction”)  Recursive enumerable but not recursive  Can return in finite steps and answer YES if α  L VALID or α  L SAT C  Can’t return in finite steps and answer NO otherwise Semi-Decidability of L VALID & L SAT C : Review

11. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence L FOL-VALID C (written L VALID C ): Language of Non-Valid Sentences Deciding Membership  Given: KB, α  Decide: KB ⊭ α? (Is there a counterexample to α? i.e., is ¬α satisfiable?) Procedure  Test whether KB  {α} ⊢ RESOLUTION   Answer YES if it does NOT L FOL-SAT (written L SAT ) Language of Satisfiable Sentences Dual Problems Undecidable: L VALID C, L SAT  RE (“Find A Counterexample”)  Not recursive enumerable  Can return in finite steps and answer NO if α  L VALID C or α  L SAT  Can’t return in finite steps and answer YES otherwise Undecidability of L VALID C & L SAT : Review

12. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Universe of Decision Problems Recursive Enumerable Languages (RE) Recursive Languages (REC) Decision Problems: Review Co-RE (RE C ) L H : Halting problem L D : Diagonal problem

13. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence What Is an Ontology, Anyway? Wilson, T. V. (2006). How Semantic Web Works. http://bit.ly/1AKeOn © 2009 Wikipedia. http://en.wikipedia.org/wiki/Ontology_(information_science)

14. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence What Are Description Logics? © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

15. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence DL Basics © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

16. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence The DL Family [1]: ALC Adapted from slides © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

17. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence The DL Family [2]: SHOIN & Web Ontology Language Based on slide © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

18. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence DL Knowledge Base Adapted from slides © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

19. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Ontologies and The Semantic Web (Web 3.0) Based on slide © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

20. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence WWW Consortium Web Ontology Language (OWL) Based on slide © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

21. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Class / Concept Constructors © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

22. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Ontology Axioms © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X

23. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Why Description Logic? OWL exploits results of 15+ years of DL research  Well defined (model theoretic) semantics  Formal properties well understood (complexity, decidability)  Known reasoning algorithms  Implemented systems (highly optimised) Adapted from slides © 2006 I. Horrocks, University of Manchester http://bit.ly/10Oh4Xhttp://bit.ly/10Oh4X “I can’t find an efficient algorithm, but neither can all these famous people.” [Garey & Johnson. Computers and Intractability: A Guide to the Theory of NP- Completeness. Freeman, 1979.]

24. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Terminology Decision Problems: True-False for Membership in Formal Language  REC (decidable) vs. RE (semi-decidable OR decidable)  Co-RE (undecidable)  Russell’s Paradox: does the barber shave himself? Ontology: Formal, Explicit Specification of Shared Conceptualization  Tells what exists (entities, objects)  Tells how entities can relate to one another  Can be used as basis for reasoning about objects, sets  Formalized using logic (e.g., description logic) Knowledge Engineering (KE): Process of KR Design, Acquisition  Knowledge  What agents possess (epistemology) that lets them reason  Basis for rational cognition, action  Knowledge gain (acquisition, learning): improvement in problem solving  Next: more on knowledge acquisition, capture, elicitation  Techniques: protocol analysis, subjective probabilities (later)

25. Computing & Information Sciences Kansas State University Lecture 16 of 42 CIS 530 / 730 Artificial Intelligence Summary Points Last Class: Resolution Theorem Proving, 9.5 (p. 275-294), R&N 2 e  Proof example in detail  Paramodulation and demodulation  Resolution strategies: unit, linear, input, set of support  FOL and computability: complements (different difficulty) and duals (same) Today: Prolog in Brief, Knowledge Engineering (KE), Ontologies  Prolog examples  Knowledge engineering  Introduction to ontologies  Ontologies defined  Ontology design  Description logics  SHOIN  Web Ontology Language (OWL) Next Class: More Ontology Design, KE; Situation Calculus Redux Coming Week: Ontologies, Description Logics, Semantic Nets